Next: References Up: Writing the SQLEddi notation Previous: Abstract Contents
The practical objective of this task is to study and understand the implementation of the SQLEddi parser that uses agent-oriented parsing techniques using empirical modelling tools. Results from the study and the experiments will introduce a new structure to the existing parser.
Based on previous experience in developing the eddi parse, the SQLEddi notation (%toysql) will be written in the aop_v204 using the notation builder tool (Harfield, 2003):
The agent script language (which determine the actions that an agent performs in the environment) will be changed to eliminate the use of difficult string substitutions ($$ and %%) and replace it with assigning only two script variable to each agent (one to communicate with its parent and one to communicate with its child), this will consequently ease the communication between agents in the agent hierarchy as well as it will make the notation easier to comprehend and develop.
The existing instructions of the script language (declare, setparas, allparas, execute and later) will be replaced with the aop_v204 execute instructions: now and later as there will be no need to declare variables in the new design (since every agent has one variable to communicate with its parent and another one to communicate with its child agent). This feature will make the SQLEddi notation (with about 100 agents) a lot shorter and simpler to understand. As a result, all SQLEddi statements will begin with the tag action instead of the old style starting with script, which means that the majority of the notation with be changed.
The modifications that will be made to the parser, does not mean changing the content of the SQLEddi environment itself (SQLEddi operations), it will only change the behaviour of the parser.
Tools to be used: In addition to the Builder, different other EM tools will be used: Dependencies in the notation will be verified and views using the DMT. The SQLEDDI environment (the uneddifying interface) will be used to study the use of SQL in conjunction with Eddi. The parsing visualisation tool will be used to visualise the parsing process.